A comparison of discrete, single stage lot-sizing heuristics with special emphasis on rules based on the marginal cost principle

Abstract

Several lot-sizing techniques suitable for use in connection with time-phased inventory records have been suggested in the last few years. The common denominator for most of these rules is their foundation in the principle of equating marginal costs per unit of time. After outlining the structures of six recent lot size models, this paper presents the results from a large simulation experiment involving a total of eleven discrete rules. It is shown that the lot-sizing techniques based on marginal costs clearly outperform the more well known rules like Part Period Balancing and Least Unit Cost, and actually rank number 2, 3, 4, 5, and 8 after the optimizing Wagner-Whitin algorithm.